Discussion:
lock daemon and nfsv4
Tigran Mkrtchyan
2010-02-08 11:43:23 UTC
Permalink
Hi ,

I see that nfs client ties to talk to lock daemon even when you mount
with v4.
My understanding was that with v4 you don't need to do that.


Regards,
Tigran.
Steve Dickson
2010-02-08 13:17:00 UTC
Permalink
Post by Tigran Mkrtchyan
Hi ,
I see that nfs client ties to talk to lock daemon even when you mount
with v4.
My understanding was that with v4 you don't need to do that.
hmm... could you please be a bit more explicit? Maybe a tshart
network trace?

steved.
Tigran Mkrtchyan
2010-02-08 13:29:13 UTC
Permalink
Oh my apologies, it's not a lock daemon.
But still something which I do not expect

Regards,
Tigran.
Post by Steve Dickson
Post by Tigran Mkrtchyan
Hi ,
I see that nfs client ties to talk to lock daemon even when you mount
with v4.
My understanding was that with v4 you don't need to do that.
hmm... could you please be a bit more explicit? Maybe a tshart
network trace?
steved.
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: mount.trace
Url: http://linux-nfs.org/pipermail/nfsv4/attachments/20100208/69fd3b19/attachment.pl
Steve Dickson
2010-02-09 15:26:18 UTC
Permalink
Post by Tigran Mkrtchyan
Oh my apologies, it's not a lock daemon.
But still something which I do not expect
Its a known bug that didn't make it into the mainstream kernel...
The details are here:
http://linux-nfs.org/pipermail/nfsv4/2009-October/011420.html

Trond, any chance of this being committed?

steved.
Trond Myklebust
2010-02-09 17:07:55 UTC
Permalink
Post by Steve Dickson
Post by Tigran Mkrtchyan
Oh my apologies, it's not a lock daemon.
But still something which I do not expect
Its a known bug that didn't make it into the mainstream kernel...
http://linux-nfs.org/pipermail/nfsv4/2009-October/011420.html
Trond, any chance of this being committed?
Chuck's RPC_TASK_SOFTCONN patch series (which first appears in
2.6.33-rc1) was supposed to fix this issue.

What kernel are you testing, Tigran?

Trond
Chuck Lever
2010-02-09 17:14:54 UTC
Permalink
Post by Trond Myklebust
Post by Steve Dickson
Post by Tigran Mkrtchyan
Oh my apologies, it's not a lock daemon.
But still something which I do not expect
Its a known bug that didn't make it into the mainstream kernel...
http://linux-nfs.org/pipermail/nfsv4/2009-October/011420.html
Trond, any chance of this being committed?
Chuck's RPC_TASK_SOFTCONN patch series (which first appears in
2.6.33-rc1) was supposed to fix this issue.
So, RPC_TASK_SOFTCONN fixes the hang that Steve reported in October.

But it doesn't change the NFSv4 callback service to stop unregistering
with the local rpcbind at start-up.
--
chuck[dot]lever[at]oracle[dot]com
Tigran Mkrtchyan
2010-02-09 19:30:30 UTC
Permalink
Post by Trond Myklebust
Post by Steve Dickson
Post by Tigran Mkrtchyan
Oh my apologies, it's not a lock daemon.
But still something which I do not expect
Its a known bug that didn't make it into the mainstream kernel...
http://linux-nfs.org/pipermail/nfsv4/2009-October/011420.html
Trond, any chance of this being committed?
Chuck's RPC_TASK_SOFTCONN patch series (which first appears in
2.6.33-rc1) was supposed to fix this issue.
Hi Trond,

I run pnfs-all-latest from git://linux-nfs.org/~bhalevy/linux-pnfs.git

77790fc45814d27a305fe52aa71099ecfb2eedde which is currently based on
2.6.33-rc6


Tigran.
Post by Trond Myklebust
What kernel are you testing, Tigran?
Trond
Loading...